UI-UG (A Unified MLLM for UI Understanding and Generation) is a multimodal large model that simultaneously supports both UI understanding and UI generation. It supports various tasks including referring, grounding, captioning and generation.